Artificial intelligence companies are escalating their efforts in statehouses to establish regulatory frameworks for AI, leading to a strategic divergence between major players. Reports indicate that Anthropic is advocating for states to impose progressively stringent safety rules, contrasting sharply with the approach favored by its competitor, OpenAI, which seeks to unify regulations nationwide.
Divergent Strategies for AI Governance
Anthropic’s strategy involves encouraging a patchwork of increasingly rigorous state laws, rather than supporting a single set of federal or uniform state mandates. Cesar Fernandez, Anthropic’s head of U.S. state and local government relations, stated in an interview that the company is “looking for legislation that meaningfully raises the bar on safety for the most capable AI systems,” arguing against simply aiming to back up the same bill across every state.
This stance contrasts with OpenAI’s approach. The firm’s top lobbyist, Chris Lehane, has termed its lobbying efforts “reverse federalism.” This term describes OpenAI’s attempts to bypass a stalled Congress by establishing a national AI structure through mirroring legislation from various states.
The divergence in strategy is notable given that Anthropic was founded by executives who left OpenAI in 2020 due to concerns over the company’s focus on safety. Despite these tensions, both companies are pushing for stronger oversight at both state and federal levels.
Industry Response to Regulatory Uncertainty
The industry is increasingly relying on states for clarity regarding AI rules, especially as Congress has shown reluctance to pass comprehensive legislation and the White House position remains inconsistent. In response to questions about OpenAI’s campaign efforts, an OpenAI spokesperson defended “reverse federalism,” asserting that it “where effective state safeguards shape national standards, helps regulators enforce the law, gives the public clearer protections, and allows developers to focus resources on safety rather than conflicting requirements.”
Anthropic highlighted its early engagement in the policy debate by pointing out that it was the sole leading AI research laboratory to endorse California’s 2025 law, which is designed to govern advanced AI models—the first such law of its kind nationwide. While OpenAI later utilized the California proposal as an example for other states without initially taking a position on it, Anthropic viewed the measure as a launching point to intensify its safety efforts.
According to Fernandez, the accelerating development of powerful AI systems was the primary catalyst behind Anthropic’s support for more ambitious proposals in New York, Illinois, and Massachusetts, noting that these successive bills contained “real safety obligations” that exceeded those of previous legislation. He emphasized that self-reporting alone is no longer adequate.
Mandating Audits and Safety Protocols
The companies have targeted different regulatory mandates. Anthropic cited its Claude Mythos model—which the company found capable of exploiting security flaws across major computer operating systems during testing—as evidence of heightened cybersecurity risks. This capability previously triggered panic within the Trump administration, leading to export controls until vulnerabilities were addressed by both Anthropic and the government.
In recent developments, OpenAI lobbyists successfully persuaded New York Governor Kathy Hochul to modify her state’s AI safety bill to align more closely with California’s requirements. However, some safety advocates noted that OpenAI subsequently supported an Illinois measure, signed into law this month by Gov. JB Pritzker. This Illinois mandate requires major AI companies to undergo annual audits of their safety plans by independent third parties—a novel requirement.
Anthropic continues to advocate for even stricter standards. In late June, the company endorsed regulations related to an economic development bond bill in Massachusetts, calling it the nation’s strongest state AI safety proposal. This proposed language mandates that leading AI firms hire independent evaluators to assess potential catastrophic risks, such as using the technology for bioweapon development, and grants the state’s attorney general the authority to enforce this mandate.
Political Lobbying Efforts
The regulatory competition extends into political campaigning. Both companies utilize super PAC networks that have invested tens of millions of dollars in political campaigns across the country. Anthropic also began directly funding legislators in California. Fernandez stated that the company supports candidates for election and re-election if their views on AI safety regulation match its core mission: “to make sure that the transition to a world with powerful AI does well for people in this country and throughout the world.”
Fernandez confirmed that Anthropic is not coordinating lobbying efforts among employees who have made personal political contributions, adding that their engagement stems from deep concern about the future of AI if adequate governmental safety policies are not enacted.
